All animal experimental procedures in this study have complied with guidelines as outlined in the “Guide for the Care and Use of Laboratory Animals” (National Research Council, USA). The animal protocols were approved by the Institutional Animal Care and Use Committee, the College of Life Sciences, Sichuan University. Briefly, 5-week-old C57BL/6 male mice of SPF grade (Beijing HFK Bioscience) were maintained in a controlled environment (12:12 light-dark cycle) with free access to food and water. After 1 week of adaptation, the mice were randomly divided into three groups: (1) control group, by which mice were fed with ANI93 chow for 28 weeks, (2) high fat diet (HFD) group, of which 65% calorie in the chow was derived from animal fat, (3) high fat diet feeding plus chlorophyllin treatment (HFD+CHL), by which the mice under HFD feeding were treated with CHL in drinking water (30 mg/L) from the 17th week for consecutive 12 weeks (n = 8–10 for each group). Calorie intake, body weight, and fasting blood glucose were monitored accordingly. The liver, blood samples, the ileum tissues, and fecal pellets were collected for analysis.

After high fat feeding for 20 weeks, the recipient mice were divided into 3 groups and subjected to fecal transplant. Group #1, mice under HFD feeding were subjected to the transplant of fecal microbes derived from control AIN93 chow, labeled as F-C. Group #2, mice under HFD feeding were transplanted with fecal microbes, derived from high fat feeding, as F-HFD. Group #3, mice under HDF feeding were subjected to the transplant of fecal microbes derived from the CHL-treated mice, labeled as F-CHL. For fecal flora transplant, fresh donor feces as mentioned above from CHL-treated mice were collected. After washing to remove the debris and CHL in the feces, about 4 × 10⁸ colony-forming unit (CFU) bacteria in 0.1 ml suspension were administered by gavage to the recipient mice. FFT was repeated 3 times per week for consecutive 6 weeks (n = 6 for each group), in a total of 26 weeks.

Liver tissues were fixed in 4% of paraformaldehyde, and hematoxylin and eosin (H&E) staining were used for histologic analysis. Liver fibrosis was determined by Masson's Trichrome staining and the sizes of fibrotic septa were quantitated via densitometry analysis. The intestinal mucus layer was measured with a periodic acid Schiff (PAS) red stain kit (ab150680, Abcam, MA, USA), following the instruction of the manufacturer. Lymphocyte infiltration in the liver was determined by immunohistochemical staining for anti-CD3 (Cat. 16669; Abcam, MA, USA). Colorimetric images for H&E and Masson's Trichrome staining were captured via Nikon eclipse Ti-U microscope (Nikon, Japan). Tight junctions in the ileum were examined by immunofluorescent staining with anti-occludin (Cat. SC5562; Santa Cruz Biotechnology, TX, USA) and all images were captured via the Leica TCS SP5 II system (Leica, Germany).

The liver and ileum tissues were homogenized in 1 ml Trizol and total RNA was extracted. The RNA was converted to cDNA via the Transcriptor First Strand cDNA Synthesis Kit (Cat. 04897030001; Roche, MA, USA). The quantitative PCR (qPCR) system contained 2 μl cDNA, 0.2 μl forward primer, and 0.2 μl reverse primer (200 nM), 5 μl MIX, and diethylpyrocarbonate (DEPC) water was added to 10 μl. The reaction was performed with a Bio-Rad machine Cfx96. The primer sequence information is listed in Table 1. The relative microRNA (mRNA) expression was normalized in the expression of RPL-19.

SYBR green-based qPCR analysis of 16S rRNA genes was used to quantitate the relative abundance of gut bacteria. Fecal microbe DNA was extracted using a stool DNA kit (Omega, China). The qPCR system contained 2 μl DNA, 0.2 μl forward primer, 0.2 μl reverse primer (200 nM), 5 μl MIX, and DEPC water was added to 10 μl. Then, it was analyzed with the Bio-Rad Cfx96 and the value was expressed as the percentage of common bacterial readouts as the internal reference. The accuracy of the qPCR-based 16 rDNA analysis was previously validated by sequencing the PCR products. The primers for bacterial qPCR analysis are listed in Table 2.

The protocols for intra-peritoneal glucose tolerance test (IPGTT) and homeostatic model assessment of insulin resistance (HOMA-IR), fasting glucose, intestinal permeability, and histological analysis were previously reported (Zhu et al., 2017). Mice were fasted for 6 or 4 h for GTTs or ITTs, respectively. Glucose (1 g/kg body weight, 20% glucose solution) or insulin (1 U/kg body weight) was injected intraperitoneally. Blood glucose was measured by a glucose meter (Accu-Chek Active, Roche, Germany) with 5 μl of blood was collected from the tip of the tail vein. Plasma insulin concentration was measured by ELISA kit (DRE30417; RB China). HOMA-IR is determined as glucose concentration x insulin concentration divided by 22.5. Limulus amebocyte extract kit (Chinese Horseshoe Crab Reagent Manufactory, Xiamen, China) was used to detect plasma lipopolysaccharide (LPS) content. Plasma TNF-alpha levels were measured by ELISA kit (Mercodia, Uppsala, Sweden; DRE30030, China). Plasma levels of low-density lipoprotein cholesterol (LDLC), high-density lipoprotein cholesterol (HDLC), total cholesterol (CHOL), total bile acids, and alanine transaminase (ALT) levels were measured by the automation instrument in the Chengdu Public Health and Clinical Medical Center.

Data recording, processing, and calculating were completed using Microsoft Excel 2003 and GraphPad Prism 5.0. Data were plotted as a mean ± SEM. Comparisons across groups were assessed using a Student's-t-test or one-way ANOVA, followed by Tukey multiple comparison testing. Statistical significance was performed as ^*P < 0.05 or ^**P < 0.01.

Mice in C57BL/6 genetic background were fed with control AIN93 chow (Control) or high fat diet (HFD) for 28 weeks. Another group of HFD mice was given chlorophyllin (CHL) in drinking water (30 mg/L) starting from the 17th week for additional 12 weeks (HFD+CHL) (Figures 1A,B). As a common feature of high fat feeding, the mice gained their body weight starting from week 4 and doubled their weight to about 40 g in 28 weeks (Figure 1C). Administration of the green pigment in drinking water significantly reduced the body weight gain by the high fat feeding group (HFD+CHL). We also measured the waist, visceral fat, and visceral fat coefficient, which were all attenuated by CHL administration as shown in Figures 1D–F. Mice normally drink 3–5 ml water, which is about 0.09–0.15 mg CHL daily. Spinach contains chlorophyll at 0.85 mg/g, which is equivalent to 0.12–0.15 g of spinach per day for the mice. For a human with 60 kg body weight, the dose of CHL would be at 0.18–0.3 g daily.

Then we extended the assessment by including other metabolic parameters. After 28 weeks of high fat feeding, plasma levels of cholesterol, triglycerides, and LDL-C in the mice were all remarkably increased (Figure 2). Conversely, the green pigment in the drinking water statistically reduced these metabolic parameters in agreement with alleviated metabolic abnormalities, such as body weight and visceral fat. Importantly, the ratio of (HDL-C)/(LDL-C) is restored by CHL treatment.

It is well-known that overweight and central obesity are tightly connected with the key features of NAFLD (Fock and Khoo, 2013; Dietrich and Hellerbrand, 2014; Milić et al., 2014). Indeed, high fat feeding for 28 weeks led to enlarged liver mass, a key feature of NAFLD in the early phase (Figure 3A). Parenchymal damage in hepatocytes as determined by the histological assessment, a crucial character of fat degeneration in hepatocytes, was also evident by HFD feeding. Hepatic steatosis was validated by Oil Red O staining. As shown, CHL in drinking water reduced the liver mass/liver index, and improved steatosis (Figure 3B). Liver injury as indicated by elevation of serum alanine transaminase (ALT) was significantly ameliorated by CHL in drinking water in agreement with improved histological findings (Figure 3C).

Persistent inflammation is a hallmark for the initial prognosis of simple steatosis into nonalcoholic steatohepatitis (NASH). Interleukin 6 (IL-6) is highly relevant to hepatic malignancy, an ominous sign of NASH (Tsuchida et al., 2018; Anstee et al., 2019; Malehmir et al., 2019), while arginase-2 is related to persistent inflammation and skewed immune responsiveness (Delyea et al., 2018). We measured the hepatic expression of inflammatory cytokines. As shown, hepatic tumor necrosis factor-α (TNF-α, IL-6, and arginase-1 at their microRNA (mRNA) levels were increased by high fat feeding (Figure 4A). As shown in our data, CHL thoroughly suppressed the hepatic expression of these inflammatory mediators, and the data is in agreement with the reduced liver injury and improved liver function (Figure 3). Indeed, infiltration of lymphocytes, as marked by CD3+ staining, was obvious in the parenchyma of liver tissue by the mice under HFD feeding, which was thoroughly suppressed by the administration of CHL (Figure 4B), in line with the reduced expression of inflammatory cytokines.

Liver fibrosis, as a consequence of hepatic injury and inflammation, is essential for wound healing and as a key marker of NASH (Enomoto et al., 2015). Masson's Trichrome staining showed that liver fibrosis is moderately appeared by high fat feeding for 28 weeks, and such fibrosis was ameliorated by the administration of CHL (Figure 5A). Expression of Type-1 alpha collagen in the liver, as measured by RT-qPCR analysis, was significantly increased by high fat feeding, but suppressed by the administration of CHL (Figures 5B–E). In a similar pattern, CHL treatment also suppressed the expression of TGF-beta1, and alpha-smooth actin, the key genes for liver fibrosis. Moreover, TIMP1, a tissue inhibitor for matrix metalloproteinases (MMPs) is also suppressed by CHL treatment, indicating that CHL may promote MMP-mediated resolution of liver fibrosis.

Type-2 diabetes (T2D) is a major driving force for hepatic steatosis. In fact, the accumulation of hepatic triglycerides in hepatocytes serves as a compensation mechanism to reduce persistent hyperglycemia, due to systemic insulin resistance. As shown in Figure 6A, fasting glucose was evidently increased in the mice of high fat feeding, which was suppressed by CHL treatment. Intraperitoneal glucose tolerance test (IPGTT) showed glucose intolerance, an indicator of insulin resistance in the mice of high fat feeding (Figures 6B,C). Likewise, the insulin tolerance test (ITT) indicated insulin resistance was built up in the mice under high fat feeding. Conversely, CHL treatment mostly resolved the resistance (Figures 6D,E). Systemic inflammation is known as a leading cause of insulin resistance (Jiao et al., 2018; Jia et al., 2020). In fact, increased plasma TNF-α levels as an indicator for systemic inflammation in the mice under HFD feeding were mostly resolved by CHL treatment (Figure 6F). Endotoxin, a prominent intestinal promoter for systemic inflammation and insulin resistance for hepatic steatosis, is elevated in the mice under HFD feeding, was suppressed by CHL treatment (Figure 6G).

Increased plasma endotoxin levels in the mice of HFD feeding, and the suppression by CHL treatment indicate that the green pigment may modulate the gut microbiome, leading to the downregulation of systemic inflammation and restoration of insulin sensitivity, which ultimately may resolve the hepatic steatosis. As shown in Figure 7, intestinal Bacteroides, the major Gram-negative microbes, were downregulated by long-term HFD feeding, which was in agreement with increased plasma endotoxin. CHL treatment thoroughly restored intestinal Bacteroides in line with the downregulation of plasma endotoxin. Firmicutes, another major gut microbe that is responsible for fermentation and production of short chain fatty acid for obesity, were increased in their abundance by HFD feeding. Conversely, CHL treatment suppressed the abundance of Firmicutes, which is in line with the reduction of body mass. Desulfovibrio, a sulfate-reducing bacterium, was increasingly presented in the gut by HFD feeding, and CHL treatment also suppressed the microbe. Similarly, epsilon-proteobacteria and enterobacteria, which were increased by the mice under HFD feeding, were suppressed by CHL treatment. Akkermansia muciniphila, a major species of intestinal symbiotic and being regarded as a marker of eubiosis, and Prevotellaceae were both downregulated by HFD feeding as previously reported in our study (Su et al., 2016). CHL treatment unequivocally restored these microbes. All these results demonstrate that the green pigment can modulate the gut microbiome, and the restored microbiota may consequently suppress NAFLD.

Balancing the gut microbiome by CHL treatment also indicates an inverse impact on intestinal innate immunity. Damaged intestinal epithelia in the mice under HFD feeding may cause intestinal leakage for increased endotoxemia. Here, we showed that the HFD-induced intestinal inflammation, as indicated by the expression of TNF-α, IL-6, and IL-1β in the ileum of the mice, was suppressed by CHL treatment (Figures 8A–C). Conversely, downregulated expression of tight junction proteins in the ileum, including ZO-1, claudin 2, and occludin by HFD feeding, was restored by CHL treatment (Figures 8D–F). Moreover, the mucus of the ileum and colon regions was examined by Periodic acid–Schiff (PAS) staining. As shown in Figures 9A,C, HFD feeding led to the damage of the mucus layer and shortening of villi, while CHL treatment reversed the gut injury. Muc2, a major mucin gene for the intestine, was suppressed by HFD feeding, in line with the damaged intestinal epithelia. CHL treatment restored Muc2 expression, presumably through reduced intestinal inflammation (Figure 9B).

To validate the causal roles of the gut microbiome in the CHL-exerted improvement of fatty liver and metabolic disorders, we applied fecal transplant. After high fat feeding for 20 weeks, the recipient mice were divided into 3 groups and subjected to fecal transplant. For fecal flora transplants, fresh donor fecal pellets were collected from the HFD mice under CHL treatment for 6 weeks. After washing, bacteria at 4 x 10⁸ CFU were administered by gavage to the recipient mice, repeating 3 times per week for consecutive 6 weeks (Figure 10A). As shown in Figure 10B, the fecal flora derived from the CHL-treated mice could suppress the gain of body weight by the recipients under HFD feeding. Metabolic parameters including plasma cholesterol, LDL-C, and triglycerides were also improved by the fecal transplant (Figures 10C–E). Similarly, fatty liver, as measured by Oil Red O staining, and liver mass, were mostly resolved by the microbes derived from the CHL treated mice (Figure 11). Liver inflammation, as the expression of TNF-α and IL-6, was suppressed by the fecal transplant, leading to the restoration of liver functions as evidenced by the reduction of plasma ALT levels (Figure 12).